Theory or Reality: The Truth of Business Processes

Businesses run on standardized processes. Without repeatable and predictable processes, an organization cannot grow while maintaining existing operations. Stable and well-defined data and processes are at the heart of sustainable scale. We encounter this during our daily lives – often without realizing it. When we place an order on Amazon, there is no team that sits down to figure out how to get your order to your house within two days. There are well-designed and highly-optimized processes that allow every order to get to every destination in as little as hours. So, what does this look like?

These days, processes are typically designed leveraging BPMN 2.0 (Business Process Model and Notation), the standard for modelling businesses processes. For process engineers and analysts, these diagrams are probably very familiar – a nod to their ubiquity in industry. BPMN 2.0 allows for the definition of activities, sequencing, decisions, accountability, handoffs and more. Some vendors have developed software to be even more granular and specialized, offering additional notation capabilities to model a process. Ideally, this is everything you need to be able to determine how your processes should operate. But how do you determine how these processes are behaving in reality?

Business processes, while being owned and managed by individuals, are typically run by transactional systems. Thinking back to our Amazon example at the top, there are undoubtedly systems that write to databases capturing all kinds of order information.

• When was the order submitted?

• Who submitted the order?

• What was ordered?

• What payment options were chosen?

• What delivery options were chosen?

• Which partner or Amazon Fulfillment Center will fulfill the order?

• What are the most efficient delivery logistics to meet SLAs?

• Which delivery partners need to be engaged?

• When was the order shipped?

• What tracking numbers are associated to the shipment?

• When was it delivered?

• Etc.

Much, if not all, of this data is relevant to the business process model. However, with the data captured in these transactional systems, it is now possible to mine the data and generate a real-time view of what is happening. This is the magic of process mining.

Process mining is a technical science that leverages a few key pieces of data to enable visibility into the execution of business processes. The three pieces of information that are most critical are as follows:

• Case ID: This is a unique identifier for every object flowing through a process. In the Amazon example, this would be each object in the shopping cart that every user in the process scope purchased… in this case this could be the orderID, or the orderItemID.

  
  

• Activity: This is the step in the process that an item will or may go through – everything from “Order Submitted” to “Shipment Notice Sent” to “Order Cancelled.” These should line up with the activities in the process model so that we can measure the actual process against the model.

  
  

• Timestamp: In a transactional database, the time at which records are created is almost always captured. This becomes critical as part of process mining because the timestamp allows us to properly order the activities.

  
  

  
  

As mentioned above, most of this data is already captured in business systems. The missing piece is the collection and transformation of data in a format that allows process owners to see how the process is occurring. While the data transformation process differs among products, most process mining systems leverage an event log to capture this information. Once the event log has been generated and populated, the data is loaded into a tool that allows you to view your business process and its actual performance. Process owners can see which process paths are most traversed, which undesired activities happen the most, when activities happen out of order, and so on. Almost any process-centric problem can at least be visualized – if not resolved – leveraging Process Mining.

The idea’s simplicity is part of its genius. The event log only requires three pieces of information and can provide a wealth of invaluable process information. Additionally, there are even more fields you can add to an event log to provide more context and information – usernames, user types, original and new values to capture changes, etc. Crucially, though, you already have this data in the systems that run your business processes. All that’s left is to leverage it to improve operational efficiency and business outcomes.